Using Divide-and-Conquer to Improve Tax Collection

Tax collection by capacity constrained governments may exhibit multiple equilibria: if delinquency is low, limited enforcement capacity is enough to discipline deviators; if delinquency is high, limited enforcement capacity is overstretched and no longer dissuasive. In principle, divide-and-conquer, a theoretically important but untested principle from mechanism design, can be used to unravel the undesirable high-delinquency equilibrium. We investigate the challenge of doing so in practice. Our preferred mechanism takes the form of Prioritized Iterative Enforcement (PIE). Tax-payers are assigned a rank trading-off expected collection and expected capacity use. Tax-payers are then iteratively threatened in small groups for which collection capacity is sufficient to induce compliance. After repayment occurs, unused collection capacity is released to issue the next round of threats. In partnership with a district of Lima (Peru) we experimentally evaluate the impact of PIE on the collection of property taxes from 13432 tax-payers. Reduced-form evidence both validates and refines the theoretical benchmark. A structural model of tax-payer behavior suggests that, keeping the number of collection actions fixed, PIE would increase tax revenue by 11.3%.